v2: Add overflow check and remove unnecessary check (Julien Cristau)
This addresses:
CVE-2017-12184 in XINERAMA
CVE-2017-12185 in MIT-SCREEN-SAVER
CVE-2017-12186 in X-Resource
CVE-2017-12187 in RENDER
A client can send a big request where the 32B "length" field has value
0. When the big request header is removed and the length corrected,
the value will underflow to 0xFFFFFFFF. Functions processing the
request later will think that the client sent much more data and may
touch memory beyond the receive buffer.
Generating strings for XKB data used a single shared static buffer,
which offered several opportunities for errors. Use a ring of
resizable buffers instead, to avoid problems when strings end up
longer than anticipated.
XkbStringText escapes non-printable characters using octal numbers.
Such escape sequence would be at most 5 characters long ("\0123"), so
it reserves 5 bytes in the buffer. Due to char->unsigned int
conversion, it would print much longer string for negative numbers.
Otherwise it can belong to a non-existing client and abort X server with
FatalError "client not in use", or overwrite existing segment of another
existing client.
stsp@ reported that modesetting(4) has been reported unreliable
on his laptop, while intel(4) works.
XXXX to be removed after 6.2 to figure out and fix the issue.
ok kettenis@, also discussed briefly with deraadt@ during EuroBSDCon.
It is supposed to be slow, and when such instructions are used to copy
data from/to mapped video memory, some hypervisors (e.g. KVM,
Microsoft Hyper-V) can generate SIGILL or SIGBUS exceptions, causing
Xorg to crash.
Bug report to OpenBSD by Max Parmer, fix from FreeBSD (Dimitry Andric)
via kettenis@
ok kettenis@
care of autoconfiguration based on the information returned by the
WSDISPLAYIO_GTYPE ioctl of the console FD. This should fix selection of
wsfb on loongson and sgi when using a non-KMS kernel driver.
ok matthieu@, jsg@
later. This matches what several Linux distros do these days as it tends to
work better than the intel driver in most cases.
There are some performance issues with vncviewer on at least Ivy Bridge and
Haswell. But for now that regression outweighs the benefits.
ok robert@, tedu@, sthen@
Recent X server doesn't require to build the DIX with different
options for xfree86 and kdrive.
Tested for beeing a no-op on m88k by Kenji Aoyama. Thanks
Completely broken on VGA cards (libpciaccess doesn't know about
the alpha specific ways of mapping the PCI space),
only XRender is broken on TGA cards.
Broadwell is still a bit weak and the modesetting driver seems to work
better than the intel driver, while still providing 3D acceleration and
video playback support.
ok phessler@, matthieu@, jsg@
HAVE_XSHMFENCE isn't defined. This makes sure the relevant "private key"
gets installed such that we don't hit a fatal assert later.
ok jsg@, matthieu@
